本文探討倒傳遞?神經網?(BPN)應用於基層?融財務危機之預測，本研究比較各預警模式之估計樣本預測能?的實證結果顯示，其正確預測能?依序為原始財務變?倒傳遞網?模式正確預測?(81.1%)為最佳模式，其他依序為因素分析後倒傳遞網?模式(77.85%)及Ordered Logit模式(75.9%)。事實上，漁會信用部這?基層?融經營?出現問題，將會引起?鎖損害與信用危機問題，故基層?融營運?需要有效的預警系統，基此，本文建議之?神經網?可提供基層?融單位及早發現問題，並採取相關的預防或管?措施。

This paper applies the Back-Propagation Network (BPN) to build the financial distress prediction models. Empirical results show that the effect of BPN on crisis management mechanisms towards communities' financial institutions in Taiwan is doing quite fine. In addition, the predictability comparison indicates that the highest accuracy is the Primitive BPN (81.1%) in the surveillance system, followed by the Factory BPN (77.85%) and the Ordered Logit (75.9%). Damages and impacts to the fishing community and industry are always far more serious when financial crises occur in the community's financial institutions. Thus, a more accurate financial warning system for governing these financial institutions is needed more than ever. The artificial neural network (ANN) suggested in this study can provide a bankruptcy predictor of financial distress among credit unions.